"What’s the story?" This question, while seemingly simple, is such an important component of what we do at Othot. Here is why.
Anytime we work with (customer) data we begin with a plan, mapping out everything that is required to accomplish the tasks we are working to complete (i.e. what data we need, which algorithms are necessary to process said data and subsequently build a model, what insights we are expecting to gain, etc.). With all of this established we move forward with execution, building out and testing the necessary scripts. With all components working properly, we can initiate the model build. Upon completion of the build, and determining through a series of metrics that the model performance is satisfactory, we’ve produced the output necessary to complete the task at hand and thus the results we promised to deliver for our customers.
We have our model and we have our output, but "what's the story?"
Do the results make sense?
Are the most influential features intuitive?
What does the distribution of probabilities look like?
Overall, does the output reflect what we anticipated to see when we first mapped out our plans? If yes, what can we learn from that? Or if no, what does that tell us?
We at Othot ask these probing questions to better define the insights gained from our models and the actions they drive, and to ultimately understand the story behind the data science. We focus on building successful models and then translating the results from these models into knowledge that we, and our customers, can learn from, solve problems with, and use to make more educated business decisions overall. So, "what’s your story?"
